A water quality identification method and device based on truncated pinball loss support vector machine

Abstract

The present invention discloses a water quality discrimination method and device based on a truncated pinball loss support vector machine, the method comprising: constructing a water quality discrimination model based on a truncated pinball loss support vector machine; optimizing the water quality discrimination model using a training sample data set to obtain a trained water quality discrimination model; inputting water quality detection data of a target water body into the trained water quality discrimination model to obtain a model output result, and determining the water quality discrimination result of the target water body according to the model output result. Constructing an optimization problem based on a truncated pinball loss support vector machine; optimizing the discrimination model using the least squares method to derive an optimized fast training model; the algorithm adopted has both feature noise robustness and sparsity, while greatly reducing the training time of the algorithm, so that the water quality detection discrimination scheme has a variety of good classification performances.

Description

Technical Field

[0001] The invention belongs to the field of support vector machines in machine learning, and relates to a water quality discrimination method and device based on a truncated pinball loss support vector machine. Background Art

[0002] Water quality testing is becoming more and more important in modern society. However, as more and more water quality testing indicators are needed, the existing water quality identification methods are not particularly accurate and efficient. Whether drinking water meets drinking standards or whether industrial wastewater meets discharge standards, in many scenarios, it is necessary to judge whether it is qualified based on water quality testing indicators. Therefore, a classification algorithm is needed to train multiple data indicators to obtain a discrimination model that can efficiently and accurately judge whether the water quality is qualified.

[0003] Support vector machine (SVM) is a linear classifier that performs binary classification on a large amount of data in a supervised learning manner. The idea is to solve the learning samples to obtain the maximum margin hyperplane, that is, the decision boundary, which can transform the problem into a problem of solving convex quadratic programming. Compared with neural networks and logistic regression, the advantage of support vector machines is that they provide a clearer and more powerful way to learn complex nonlinear equations.

[0004] Specifically, when the samples are linearly separable, the optimal classification hyperplane for the two types of samples is found in the original space. When the samples are not linearly separable, slack variables are added and the samples in the low-dimensional input space are mapped to the high-dimensional space using nonlinear mapping to make them linearly separable, so that the optimal classification hyperplane can be found in the feature space.

[0005] After obtaining the optimal hyperplane for classifying sample data, the training test data can be tested. When the correct classification rate reaches a certain level, the unknown data can be classified and predicted. This is the original and most direct application of SVM.

[0006] Specifically, the traditional hinge loss support vector machine (C-SVM) uses hinge loss to maximize the width of the interval between the closest points of different classes. Since points that are fully correctly classified are not penalized, C-SVM usually has good sparsity, that is, the percentage of support vectors is low. For classification problems, the noise in the input is called feature noise. Usually, feature noise causes perturbations around the boundaries of different classes. Since the final decision hyperplane is determined by a relatively small number of support vectors, C-SVM is sensitive to feature noise and is unstable to resampling.

[0007] Later, someone chose to use the pinball loss function instead of the hinge loss function, and derived the pinball loss support vector machine (pin-SVM). Unlike C-SVM, pin-SVM penalizes many correctly classified points. Although pin-SVM is noise robust, it completely loses sparsity. The lack of sparsity is usually fatal to SVM models because it affects the computation time, especially in the prediction stage. Summary of the invention

[0008] Objective: To overcome the deficiencies in the prior art, the present invention provides a water quality identification method and device based on a truncated pinball loss support vector machine, so that the method has both noise robustness and sparsity, and can greatly reduce the training time of the algorithm.

[0009] Truncated Pinball Loss Support Vector Machine A new loss function is proposed, which combines the advantages of the hinge loss function and the pinball loss function. Since the horizontal part in the loss function curve is related to sparsity, the present invention chooses to flatten the negative part of the pinball loss function at an appropriate position, and penalizes the correctly classified data to make it insensitive to feature noise.

[0010] Least Squares Support Vector Machine (LS-SVM) is a machine learning algorithm. As an improved support vector machine based on statistical theory, LS-SVM has an advanced and complete theoretical system, which can transform the solution of quadratic optimization problems into the solution of linear equations, thereby simplifying the solution of the problem. However, due to the introduction of the square term of the error and the equality constraint in the objective function, although the solution method is simplified, the final decision function is related to all samples, that is, the excellent property of sparsity is lost, and it is sensitive to noise and unstable in resampling. At present, in order to avoid the influence of this undesirable characteristic, a "pruning" algorithm can be used to achieve the sparsity of the solution to a certain extent.

[0011] Technical solution: To solve the above technical problems, the technical solution adopted by the present invention is:

[0012] In a first aspect, a water quality identification method based on a truncated pinball loss support vector machine is provided, comprising:

[0013] Construct a water quality discrimination model based on truncated pinball loss support vector machine;

[0014] Optimizing the water quality discrimination model using a training sample data set to obtain a trained water quality discrimination model;

[0015] The water quality detection data of the target water body is input into the trained water quality discrimination model to obtain the model output result, and the water quality discrimination result of the target water body is determined according to the model output result.

[0016] In some embodiments, inputting the water quality detection data of the target water body into the trained water quality discrimination model includes:

[0017] The water quality test data is normalized according to a preset normalization method, and then the normalized data is mapped from the original space to a higher-dimensional feature space using a linear kernel function to obtain the preprocessed water quality test data, which is then input into the water quality discrimination model.

[0018] In some embodiments, the training sample data set includes multiple sets of water quality detection data with category labels, wherein the category labels include qualified and unqualified.

[0019] In some embodiments, the discriminant function of the water quality discriminant model is:

[0020]

[0021] Wherein, sign(·) is the sign function, l represents the number of samples in the training sample data set, and x i represents the i-th sample in the sample set, α i Represents x i The weight vector of the model, b is the bias value of the model parameters, y i represents the discrimination result corresponding to the i-th sample, y(x) is the result of the discriminant function, φ(x,x i ) is x and x i After mapping to the feature space, if y(x) is 1, the target object identification result is qualified; if y(x) is -1, the target object identification result is unqualified.

[0022] In some embodiments, the water quality discrimination model is optimized using a training sample data set to obtain a trained water quality discrimination model, including:

[0023] Step 1: Obtain a normalized training sample dataset of water quality detection data where x i ∈R n And y i ∈{-1,1};

[0024] Step 2: Using the linear kernel function Map the samples from the original space to a higher-dimensional feature space to obtain the processed training sample data set, that is,

[0025] Step 3: Use the processed training sample data set to solve the pre-constructed linear equations, calculate the optimized variables, and obtain the trained water quality discrimination model.

[0026] In some embodiments, the linear equations include:

[0027]

[0028] Where ω is the normal vector of the classification hyperplane, ω T is the transposed matrix of ω, l represents the number of samples in the sample set, x i represents the i-th sample in the sample set, α i is the Lagrange multiplier, which means x i The weight vector, y i represents the discrimination result corresponding to the i-th sample, φ(x i ) represents the feature vector that maps the sample from the original space to the high-dimensional space, b is the deviation value of the model parameter, ξ i is the slack variable, C is the penalty coefficient, τ and s are constants, and are adjustable parameters of the model.

[0029] Furthermore, the method for constructing the linear equation system includes:

[0030] Constructing a support vector machine based on truncated pinball loss The optimization problem P1 is:

[0031]

[0032] in P τ,s (u) represents the truncated pinball loss function, H τ (u) is an auxiliary function introduced, u is an intermediate parameter, Indicates that u is defined in the real number range, τ and s are constants, which are adjustable parameters of the model, and 0≤τ≤1, ω is the normal vector of the classification hyperplane, ω T is the transposed matrix of ω, C is the penalty coefficient, y i represents the discrimination result corresponding to the i-th sample, φ(x i ) represents the feature vector that maps the sample from the original space to the high-dimensional space, x i represents the i-th sample in the sample set, and b is the deviation value of the model parameter;

[0033] By introducing slack variables ξ=(ξ1,ξ2,…ξ l ) T , the optimization problem P1 is transformed into the optimization problem P2:

[0034]

[0035] in ξ i is the slack variable;

[0036] The optimization problem P2 is transformed into the optimization problem P3 using the least squares form:

[0037]

[0038] in

[0039] Construct the Lagrangian function L(ω,b,ξ) corresponding to the optimization problem P3 i ,α i ):

[0040]

[0041] where α i is the Lagrange multiplier, which means x i The weight vector of the optimization problem P3 is obtained by i , α i The derivative of is set to zero to obtain the linear equation system; by solving the linear equation system, the optimization variables b, α are obtained. i , substitute into the discriminant function of the water quality discriminant model to obtain the water quality discriminant model.

[0042] In a second aspect, the present invention provides a water quality identification device based on a truncated pinball loss support vector machine, including a processor and a storage medium;

[0043] The storage medium is used to store instructions;

[0044] The processor is used to operate according to the instructions to execute the steps of the method according to the first aspect.

[0045] In a third aspect, the present invention provides a storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method described in the first aspect.

[0046] Beneficial effects: The water quality identification method and device based on truncated pinball loss support vector machine provided by the present invention have the following advantages: The present invention improves the shortcomings of C-SVM that is sensitive to feature noise and unstable resampling, and improves The disadvantages of high training time complexity and long training time are improved by the poor sparsity caused by LS-SVM treating almost all training data as support vectors. The improved algorithm has both noise robustness and sparsity, and can greatly reduce the training time of the algorithm. In order to overcome the shortcomings of long training time and high time complexity of data sets, the present invention combines the idea of ​​LS-SVM and adopts the least squares linear system as the loss function, avoiding the need for existing support vector machines to solve quadratic programming problems, thereby improving the speed of solving problems. The discriminant model is optimized, and the algorithm adopted has both feature noise robustness and sparsity, while greatly reducing the training time of the algorithm, so that the water quality detection discrimination scheme has a variety of good classification performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A schematic diagram of a water quality identification method according to an embodiment of the present invention;

[0048] Figure 2 The figure is a training flowchart of an improved support vector machine algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The present invention will be further described below in conjunction with the accompanying drawings and embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0050] In the description of the present invention, "several" means more than one, "many" means more than two, "greater than", "less than", "exceed", etc. are understood to exclude the number itself, and "above", "below", "within", etc. are understood to include the number itself. If there is a description of "first" or "second", it is only used for the purpose of distinguishing the technical features, and cannot be understood as indicating or implying the relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0051] In the description of the present invention, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0052] Example 1

[0053] A water quality identification method based on truncated pinball loss support vector machine, comprising:

[0054] Construct a water quality discrimination model based on truncated pinball loss support vector machine;

[0055] Optimizing the water quality discrimination model using a training sample data set to obtain a trained water quality discrimination model;

[0056] The water quality detection data of the target water body is input into the trained water quality discrimination model to obtain the model output result, and the water quality discrimination result of the target water body is determined according to the model output result.

[0057] In some embodiments, inputting the water quality detection data of the target water body into the trained water quality discrimination model includes:

[0058] The water quality test data is normalized according to a preset normalization method, and then the normalized data is mapped from the original space to a higher-dimensional feature space using a linear kernel function to obtain the preprocessed water quality test data, which is then input into the water quality discrimination model.

[0059] In some embodiments, the training sample data set includes multiple sets of water quality detection data with category labels, wherein the category labels include qualified and unqualified.

[0060] In some embodiments, the discriminant function of the water quality discriminant model is:

[0061]

[0062] Wherein, sign(·) is the sign function, l represents the number of samples in the training sample data set, and x i represents the i-th sample in the sample set, α i Represents x i The weight vector of the model, b is the bias value of the model parameters, y i represents the discrimination result corresponding to the i-th sample, y(x) is the result of the discriminant function, φ(x,x i ) is x and x i After mapping to the feature space, if y(x) is 1, the target object identification result is qualified; if y(x) is -1, the target object identification result is unqualified.

[0063] In some embodiments, the water quality discrimination model is optimized using a training sample data set to obtain a trained water quality discrimination model, including:

[0064] Step 1: Obtain a normalized training sample dataset of water quality detection data where x i ∈R n And y i ∈{-1,1};

[0065] Step 2: Using the linear kernel function Map the samples from the original space to a higher-dimensional feature space to obtain the processed training sample data set, that is,

[0066] Step 3: Use the processed training sample data set to solve the pre-constructed linear equations, calculate the optimized variables, and obtain the trained water quality discrimination model.

[0067] In some embodiments, a method for constructing a water quality discrimination model based on a truncated pinball loss support vector machine includes:

[0068] Constructing a support vector machine based on truncated pinball loss Optimization problem;

[0069] Through The objective function is optimized and the optimized linear equations are derived;

[0070] The processed training sample data set is used to solve the pre-constructed linear equations and calculate the optimized variables to determine the trained water quality discrimination model.

[0071] In the implementation process of this embodiment, firstly construct After obtaining the optimization model, this embodiment further improves the optimization model. In particular, this embodiment uses the least squares method to Optimize to obtain a training model represented by a set of linear equations, input unknown water quality test data into the optimized training model to obtain a discriminant model. In addition, a training sample set is also required in the optimization process, which includes a variety of water quality test data and corresponding classification labels. Finally, the unknown water quality test data can be input into the optimized discriminant model to obtain the model output result, and determine whether the unknown water quality is qualified based on the model output result.

[0072] During the implementation process, The objective function is optimized and the optimized fast training model is derived. The optimization objective of the truncated pinball support vector machine is shown in the optimization problem P1. This paper selects a more advanced support vector machine algorithm with many excellent properties, and this paper makes improvements on this basis.

[0073] Will The constraint condition uses equality constraints to replace the original inequality constraints, and the slack variable ξ in the objective function is i Replaced with ξ i 2 , omitting the slack variable ξ iThe difference between these two aspects also leads to the difference between the two methods in the solution process. The original method needs to go through multiple iterations until the termination condition is met, while the method of the present invention only needs to solve a linear equation system, and the solution is obviously faster.

[0074] The method for constructing the linear equation system includes:

[0075] Constructing a support vector machine based on truncated pinball loss The optimization problem P1 is:

[0076]

[0077] in P τ,s (u) represents the truncated pinball loss function, H τ (u) is an auxiliary function introduced, u is an intermediate parameter, Indicates that u is defined in the real number range, τ and s are constants, which are adjustable parameters of the model, and 0≤τ≤1, ω is the normal vector of the classification hyperplane, ω T is the transposed matrix of ω, C is the penalty coefficient, y i represents the discrimination result corresponding to the i-th sample, φ(x i ) represents the feature vector that maps the sample from the original space to the high-dimensional space, x i represents the i-th sample in the sample set, and b is the deviation value of the model parameter;

[0078] By introducing slack variables ξ=(ξ1,ξ2,…ξ l ) T , the optimization problem P1 is transformed into the optimization problem P2:

[0079]

[0080] in ω is the normal vector of the classification hyperplane, C is the penalty coefficient, ξ i is the slack variable, y i represents the discrimination result corresponding to the i-th sample, x i represents the i-th sample in the sample set, φ(x i ) represents the feature vector that maps the sample from the original space to the high-dimensional space, and b is the deviation value of the model parameter;

[0081] The optimization problem P2 is transformed into the optimization problem P3 using the least squares form:

[0082]

[0083] in ω is the normal vector of the classification hyperplane, C is the penalty coefficient, ξi is the slack variable, x i represents the i-th sample in the sample set, φ(x i ) represents the feature vector that maps the sample from the original space to the high-dimensional space, y i represents the discrimination result corresponding to the i-th sample, b is the deviation value of the model parameter, τ and s are constants and are adjustable parameters of the model;

[0084] Construct the Lagrangian function L(ω,b,ξ) corresponding to the optimization problem P3 i ,α i ):

[0085]

[0086] where α i is the Lagrange multiplier, ω is the normal vector of the classification hyperplane, C is the penalty coefficient, ξ i is the slack variable, φ(x i ) represents the feature vector that maps the sample from the original space to the high-dimensional space, x i represents the i-th sample in the sample set, b is the deviation value of the model parameter, τ and s are constants, and are the adjustable parameters of the model. By replacing ω, b, ξ in the Lagrangian function corresponding to the optimization problem P3 i , α i The derivative of is set to zero, resulting in a system of linear equations;

[0087]

[0088] where α i is the Lagrange multiplier, which means x i The weight vector, x i represents the i-th sample in the sample set, ω is the normal vector of the classification hyperplane, y i represents the discrimination result corresponding to the i-th sample, l represents the number of samples in the sample set, b is the deviation value of the model parameter, ξ i is the slack variable, C is the penalty coefficient, φ(x i ) represents the feature vector that maps the sample from the original space to the high-dimensional space, τ and s are constants and are adjustable parameters of the model; by solving the above linear equations, we can get the optimization variables b and α i , substitute the discriminant function of the water quality discriminant model to obtain the water quality discriminant model. This solution is simpler and faster than solving the quadratic programming problem.

[0089] This optimization method is similar to C-SVM or The difference is that the constraint condition of this optimization method adopts equality constraint, which replaces the inequality constraint of the latter. The second is to replace the slack variable ξ in the objective function. i Replaced with And the slack variable ξ is omitted i Finally, the problem is further simplified by solving the equality constraint and the least squares problem.

[0090] Step 1: Select the training data set of the collected water quality test data where x i ∈R n And y i ∈{-1,1}, normalize the data set.

[0091] Step 2: Select a linear kernel function Map the samples from the original space to a higher-dimensional feature space, that is,

[0092] Step 3: Substitute the training data set into the derived training model;

[0093] Step 4: Input the training model in step 3 into MATLAB for calculation.

[0094] Step 5: The discriminant function of the support vector machine discriminant model is

[0095] In summary, the present embodiment provides a water quality identification method based on a truncated pinball loss support vector machine, comprising: constructing a water quality identification method based on a truncated pinball loss support vector machine. optimization problem; by The objective function is optimized to derive an optimized fast training model; the training model is used to train the water quality sample data set to determine the discriminant model for the sample; the various indicators of the collected water quality are input into the discriminant model to obtain the output of the model, and the water quality is judged according to the output of the model. The discriminant model is optimized, and the algorithm adopted has both feature noise robustness and sparsity, while greatly reducing the training time of the algorithm, so that the water quality detection discrimination scheme has a variety of good classification performance.

[0096] Example 2

[0097] In a second aspect, the present embodiment provides a water quality identification device based on a truncated pinball loss support vector machine, including a processor and a storage medium;

[0098] The storage medium is used to store instructions;

[0099] The processor is used to operate according to the instructions to execute the steps of the method according to embodiment 1.

[0100] Example 3

[0101] In a third aspect, this embodiment provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in Example 1 are implemented.

[0102] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0103] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0104] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0105] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0106] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A water quality identification method based on truncated pinball loss support vector machine, characterized in that: include: A water quality discrimination model based on truncated pinball loss support vector machine is constructed; wherein the discrimination function of the water quality discrimination model is: Among them, sign(·) is the sign function, l represents the number of samples in the training sample data set, and x i represents the i-th sample in the sample set, α i Represents x i The weight vector of the model, b is the bias value of the model parameters, y i represents the discrimination result corresponding to the i-th sample, y(x) is the result of the discriminant function, φ(x,x i ) is x and x i After mapping to the feature space, if y(x) is 1, the target object is qualified; if y(x) is -1, the target object is unqualified. The water quality discrimination model is optimized using the training sample data set to obtain a trained water quality discrimination model, which specifically includes: Step 1: Obtain a normalized training sample data set of water quality detection data; Step 2: Use the linear kernel function to map the samples from the original space to a higher-dimensional feature space to obtain the processed training sample data set; Step 3: Solve the pre-constructed linear equations using the processed training sample data set, calculate the optimized variables, and obtain a trained water quality discrimination model; wherein the linear equations include: Where ω is the normal vector of the classification hyperplane, ω T is the transposed matrix of ω, α i is the Lagrange multiplier, which means x i The weight vector, y i represents the discrimination result corresponding to the i-th sample, φ(x i ) represents the feature vector that maps the sample from the original space to the high-dimensional space, b is the deviation value of the model parameter, ξ i is the slack variable, C is the penalty coefficient, τ is a constant, and is an adjustable parameter of the model; The method for constructing the linear equation system includes: Constructing a support vector machine based on truncated pinball loss The optimization problem P1 is: in P τ,s (u) represents the truncated pinball loss function, H τ (u) is an auxiliary function introduced, u is an intermediate parameter, Indicates that u is defined in the real number range, τ and s are constants, which are adjustable parameters of the model, and 0≤τ≤1, ω is the normal vector of the classification hyperplane, ω T is the transposed matrix of ω, C is the penalty coefficient, y i represents the discrimination result corresponding to the i-th sample, φ(x i ) represents the feature vector that maps the sample from the original space to the high-dimensional space, and b is the deviation value of the model parameter; By introducing slack variables ξ=(ξ1,ξ2,…ξl) T , the optimization problem P1 is transformed into the optimization problem P2: in ξ i is the slack variable; The optimization problem P2 is transformed into the optimization problem P3 using the least squares form: in Construct the Lagrangian function L(ω,b,ξ) corresponding to the optimization problem P3 i ,α i ): where α i is the Lagrange multiplier, which means x i The weight vector of the optimization problem P3 is obtained by i , α i The derivative of is set to zero to obtain the linear equation system; by solving the linear equation system, the optimization variables b, α are obtained. i , substitute into the discriminant function of the water quality discriminant model to obtain the water quality discriminant model; The water quality detection data of the target water body is input into the trained water quality discrimination model to obtain the model output result, and the water quality discrimination result of the target water body is determined according to the model output result.

2. The water quality identification method based on truncated pinball loss support vector machine according to claim 1 is characterized in that: Input the water quality detection data of the target water body into the trained water quality discrimination model, including: The water quality test data is normalized according to a preset normalization method, and then the normalized data is mapped from the original space to a higher-dimensional feature space using a linear kernel function to obtain the preprocessed water quality test data, which is then input into the water quality discrimination model.

3. The water quality identification method based on truncated pinball loss support vector machine according to claim 1 is characterized in that: The training sample data set includes multiple groups of water quality detection data with category labels, wherein the category labels include qualified and unqualified.

4. A water quality identification device based on truncated pinball loss support vector machine, characterized in that: including processor and storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1 to 3.

5. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 3 are implemented.

Images